Skip to main content

Tools for JAX.

Project description

This repository implements a variety of tools for the differential programming library JAX.

Major components

Tjax’s major components are:

  • A dataclass and mypy_plugin decorator dataclasss that facilitates defining structured JAX objects (so-called “pytrees”), which benefits from:

    • the ability to mark fields as static (not available in chex.dataclass),

    • a MyPy plugin, and

    • a display method that produces formatted text according to the tree structure.

  • A fixed_point finding library heavily based on fax. Our library

    • supports stochastic iterated functions, and

    • uses dataclasses instead of closures to avoid leaking JAX tracers.

  • A shim for the gradient transformation library optax that supports:

    • easy differentiation and vectorization of “gradient transformation” (learning rule) parameters,

    • gradient transformation objects that can be passed dynamically to jitted functions, and

    • generic type annotations.

Minor components

Tjax also includes:

  • A pretty printer print_generic for aggregate and vector types, including dataclasses. (See display.)

  • Versions of custom_vjp and custom_jvp that support being used on methods. (See shims.)

  • Tools for working with cotangents. (See cotangent_tools.)

  • A random number generator class Generator. (See generator.)

  • JAX tree registration for NetworkX graph types. (See graph.)

  • Leaky integration leaky_integrate and Ornstein-Uhlenbeck process iteration diffused_leaky_integrate. (See leaky_integral.)

  • An improved version of jax.tree_util.Partial. (See partial.)

  • A Matplotlib trajectory plotter PlottableTrajectory. (See plottable_trajectory.)

  • A testing function assert_tree_allclose that automatically produces testing code. And, a related function tree_allclose. (See testing.)

  • Basic tools like divide_where. (See tools.)

Also, see the documentation.

Contribution guidelines

  • Conventions: PEP8.

  • How to run tests: pytest .

  • How to clean the source:

    • isort tjax

    • pylint tjax

    • mypy tjax

    • flake8 tjax

Project details


Release history Release notifications | RSS feed

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

tjax-0.15.0.tar.gz (39.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

tjax-0.15.0-py3-none-any.whl (48.9 kB view details)

Uploaded Python 3

File details

Details for the file tjax-0.15.0.tar.gz.

File metadata

  • Download URL: tjax-0.15.0.tar.gz
  • Upload date:
  • Size: 39.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.1.11 CPython/3.9.6 Darwin/21.1.0

File hashes

Hashes for tjax-0.15.0.tar.gz
Algorithm Hash digest
SHA256 1b82a87835bd6ded5f565f2804d43d7d8eba8a88b2e16ffbf4f40ba5eaee67b0
MD5 81d003cf58a0579e9f6337eaab53583f
BLAKE2b-256 9c0b7073a7732da66da46a275504d9fd8ca8f46d441192f94ad6a89640307bd9

See more details on using hashes here.

File details

Details for the file tjax-0.15.0-py3-none-any.whl.

File metadata

  • Download URL: tjax-0.15.0-py3-none-any.whl
  • Upload date:
  • Size: 48.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.1.11 CPython/3.9.6 Darwin/21.1.0

File hashes

Hashes for tjax-0.15.0-py3-none-any.whl
Algorithm Hash digest
SHA256 0ada17739b1f29b22431d03747d0215cdc4a1c989e8b8f20eb717e990322604b
MD5 0b8d2fe9c46131c60e3e8f6a6369eede
BLAKE2b-256 ebd20a0f8816b9ab74f3bd117e5f19f1678b51eebb26277442c79a3d8529bf6b

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page